Atomic‐Layer‐Deposited Al₂O₃ as Effective Barrier against the Diffusion of Hydrogen from SiN_x:H Layers into Crystalline Silicon during Rapid Thermal Annealing

Abstract

Atomic-layer-deposited (ALD) Al 2 O 3 films with a thickness of a few nanometers have been successfully applied in microelectronics and photovoltaics. [1-5] In particular, in silicon-based solar cells, the introduction of Al 2 O 3 surface passivation layers was a crucial step towards higher efficiencies of industrial solar cells in recent years. The metal contacts in today's industrial silicon solar cells are made by screen-printing of metal pastes in combination with a subsequent rapid thermal annealing (RTA) step at set-peak temperatures in the range between 750 and 850 C for a few seconds. [6] To preserve the excellent passivation quality of the ALD-Al 2 O 3 layers on the silicon surface during the RTA step, the Al 2 O 3 layers are capped by silicon nitride (SiN x) layers. These top layers are grown by means of plasmaenhanced chemical vapor deposition (PECVD), resulting in amorphous SiN x :H layer with very high hydrogen content (typically in the range of 10-20 at%). [7] In contrast, ALD-Al 2 O 3 layers have a hydrogen content in the range of only 1-2 at%. [3] During the RTA step, hydrogen partly diffuses from the hydrogen-rich SiN x layer [8] through the Al 2 O 3 layer to the interface and also into the crystalline silicon bulk, where it is able to passivate defects. [9-11] Interestingly, the hydrogen was also found to be able to create new recombination centers in the silicon bulk, in some cases leading to a severe degradation in solar cell efficiency during illumination. [12-16] Therefore, in photovoltaics, the control of the amount of hydrogen diffusing into the crystalline silicon bulk has turned out to be of utmost importance. There have been conjectures in the literature that Al 2 O 3 layers might severely hamper the in-diffusion of hydrogen from SiN x :H into the silicon bulk. [17,18] However, these studies did not provide any quantitative measurements on how effective Al 2 O 3 actually is as a hydrogen barrier. This letter aims at closing this gap by quantifying the amount of hydrogen diffused into the silicon bulk through Al 2 O 3 layers of different thicknesses (5À25 nm) at varying RTA peak temperatures ϑ peak. Figure 1 shows exemplary measurements for three group A samples (see Experimental Section) with SiN x :H films of different compositions, fired at a measured RTA peak temperature of (792 AE 10) C. Directly after RTA, the hydrogen is mainly present in the form of hydrogen dimers H 2 in the silicon bulk. [19] These H 2 dimers dissociate during low-temperature annealing in darkness (e.g., at 160 C) and the hydrogen atoms subsequently passivate boron dopant atoms. [20] As a consequence, the bulk resistivity ρ of the sample increases as a function of time during dark annealing at 160 C on a hotplate. We measure the resistivities in-between the periods of 160 C-dark annealing using

On the Role of AlOxThickness in AlOx/SiNy: H Layer Stacks Regarding Light- and Elevated Temperature-Induced Degradation and Hydrogen Diffusion in c-Si

“…According to theories of diffusion based upon kinetic theory, the diffusion coefficient D is inversely proportional to the square root of the mass of the diffusing species [56]. Therefore, with L ∝ √ D, the expected ratio of the hydrogen diffusion length to that of deuterium is 4 √ 2 ≈ 1.19, which explains the lower diffusion length determined in this work compared to L [H] established by Helmich et al [40].…”

“…Interestingly, the fit yields an offset value of [D] offset = (4.2 ± 0.2) × 10 12 cm −2 . Helmich et al [40] attributed this non-zero saturation (which was lower in their work) to a hydrogen content in the AlO x layer itself diffusing into the c-Si bulk during the firing step. As the ALD AlO x layers deposited in our work were not intentionally deuterated and the natural isotope ratio for deuterium is only at 0.015%, the here-observed offset cannot be attributed to the effect described in [40].…”

“…So far, all investigations regarding hydrogen diffusion through AlO x layers were done by indirect measurements or evidence [34], [40]. SIMS measurements after the firing step give direct insight into the barrier effect of the ALD AlO x layer regarding deuterium diffusion.…”

“…Helmich et al [40] determined diffusion lengths of hydrogen in ALD AlO x by bulk resistivity measurements (here referred to as L [H] ) at different firing sample peak temperatures. From interpolation, a diffusion length of approx.…”

“…These findings could be explained by AlO x acting as gas diffusion barrier [36] and thus reducing the in-diffusion of hydrogen in the c-Si bulk, which could also explain the effects observed on regeneration kinetics of the BO-related defect [37]. Based upon the works on the formation and dissociation of hydrogen dimers in silicon of Voronkov et al [38] and subsequent studies [39], Helmich et al [40] investigated the diffusion of hydrogen through the AlO x layer by determining the boron-hydrogen concentration [BH] in the c-Si bulk. By varying the AlO x thickness, they were able to indirectly identify diffusion lengths of hydrogen in the AlO x layer being dependent on the firing peak temperature.…”